Romona -
Your F-contrast is correct for testing any difference in response shapes
between the two groups, where that shape can be captured by your three
basis functions. Daniel's suggestion of an FIR will capture an even
greater range of shape differences, though possibly at reduced power if
those differences are within the subspace defined by the canonical and
its two partial derivatives (though the nice thing about an FIR at the
2nd-level is that you can still create more focussed, ie potentially
more sensitive, F-contrasts akin to the canonical and its derivatives -
see Chapter 30 of SPM5 manual, page 248 onwards).
HOWEVER, tests of shape differences (vs baseline) require that you can
estimate the HRF shape efficiently, which means that you must have a
jittered SOA (eg of order 1-12s), eg through appreciable ration of null
events. There is no point trying to estimate response shapes (vs
baseline) if you have a short SOA design.
Rik
> Dear Ramona,
>
> Have you considered using a finite impulse response (FIR) model to analyze your event-related data? Such a model would allow you to estimate the average stimulus-locked BOLD response to each of your trial types across time. Comparisons of these response shapes for your 2 subject groups might then be more easily (i.e., visually) interpretable.
>
> Hope this helps,
> Daniel
>
> Daniel Weissman, Ph.D.
> Assistant Professor
> Department of Psychology
> University of Michigan
> Room 4052
> 1012 East Hall
> 530 Church Street
> Ann Arbor, MI 48109
> ________________________________________
> From: SPM (Statistical Parametric Mapping) [[log in to unmask]] On Behalf Of Liliana Demenescu [[log in to unmask]]
> Sent: Friday, March 27, 2009 11:22 AM
> To: [log in to unmask]
> Subject: [SPM] Hemodynamic response shape testing.
>
> Dear all,
>
> I have an event-related design. The 1st level included the canonical HRF and its time and dispersion derivatives, because we want to look if there are difference in response shape between 2 groups. At the 1st level analysis, I defined a t-contrast for each basis function (HRF, TD and DD). The 2nd level analysis employed an ANOVA 3 (basis functions) x 2 (groups). I did an F test defined as:
> 1 -1 0 0 0 0
> 0 0 1 -1 0 0
> 0 0 0 0 1 -1
> (group1 - group2: HRF; group1 - group2: time der. and group1 - group2: dispersion der.). This tells me which voxels have different activity between the 2 groups with respect to either amplitude, either time, or dispersion. I plotted the results, but I found it difficult to understand/explain, as it's an F test. Moreover, as I understood from Friston's paper (Friston et al. 1998, Neuroimage, 7:30-40) this tells me how well the model fits or have I misunderstood the things?
>
> My question is how can I look for differences in response shapes between 2 groups combining the 3 basis functions?
>
> Any suggestion is more than welcome.
>
> Thank you!
>
> Kind regards,
> Ramona
>
--
-------------------------------------------------------
DR RICHARD HENSON
MRC Cognition & Brain Sciences Unit
15 Chaucer Road
Cambridge, CB2 7EF
England
EMAIL: [log in to unmask]
URL: http://www.mrc-cbu.cam.ac.uk/people/rik.henson/personal
TEL +44 (0)1223 355 294 x522
FAX +44 (0)1223 359 062
MOB +44 (0)794 1377 345
-------------------------------------------------------
|